An enduring dispute in Late Pleistocene and Holocene archaeology of mainland Southeast Asia is the nature of the transition from forager economies to agricultural economies. As a key milestone in complex human-environment interactions, the debate has many dimensions. At one extreme of the debate is the claim that agricultural technologies and cultures appeared in Southeast Asia as a result of influence from north Asia, via the lower Yangtze River and the Yellow River. At the other extreme is the claim is that agriculture emerged from a locally contingent trajectory of changes in human-environment relationships (cf. Hunt and Rabett, 2014). While the cultivation of rice and the domestication of pigs and cattle took place in the Yangtze Valley earlier than elsewhere in mainland SEA (Chi and Hung, 2010; Higham et al., 2011), the influence of local contingencies remains poorly understood. One of the enduring challenges is that a critical period of time for this transition – the Late Pleistocene through to the middle Holocene – is sparsely represented in the archaeological record. We have a rich and well-documented record for the later Holocene when people were living more sedentary lifestyles, for example at Khok Phanom Di in Thailand and Man Bac in Vietnam. And we have many cave and rockshelter sites representing Pleistocene forager lifestyles, such as Tham Lod in Thailand and Xom Trai in Vietnam.
However, middle Holocene (c. 6000–3500 BP), the archaeological record in mainland SEA is particularly sparse. This major gap in archaeological evidence for the region has been called ‘the missing millennia’ (White and Bouasisengpaseuth, 2008:39). It is an important period because major changes occurred during this time. Ceramics appeared in many parts of Southeast Asia; domesticated plants such as millet and rice appeared; stone artefact technologies transitioned from mostly flaked to mostly ground stone artefacts; settlements expanded from primarily karstic upland and estuarine landscapes during the early Holocene to include inland alluvial lowland villages by the late Holocene (White, 2011). But the current sparsity of the archaeological record means that questions of the timing and character of these changes remain difficult to answer.
In this paper we present evidence of human activity from coastal Thailand that spans ‘the missing millennia’. Khao Toh Chong rockshelter is significant because it has a rich faunal record spanning the middle Holocene, and is located in an area with a well-documented sea level curve. This provides a unique opportunity to investigate locally contingent factors such as sea level changes on human subsistence behaviours at a critical time in the period of the transition from forager economies to agricultural economies. We report on a geoarchaeological analysis of the site to provide a local environmental context to the human occupation, as well as helping to understand site formation processes and artefact taphonomy.
The Guangxi Province of southern China has extensive evidence of a forager economy with a semi-sedentary lifestyle during c. 7-4 k BP (Higham, 2013). Cave occupation continues until 6000 BC in Xianrendong and 5000–4000 BC in Zengpiyan, and more than 30 open sites containing shell middens have been found on the terraces of the Zuojiang, Youjiang and Yongjiang rivers near Nanning, in southern Guangxi (Chi and Hung, 2012; Fu; 2002). Occupation of these sites, characterized by the largest, Dingsishan, spans 7000-3500 BC. The sites include pottery manufacturing workshops, cemeteries and large quantities of aquatic and terrestrial animal bones, indicating that fishing and hunting were important activities (no cultivars have been recovered). The archaeology of this region gives the impression of a continuous sequence of human occupation. We see gradual, overlapping adaptations resulting in changes in landscape use, the appearance of pottery and use of cemeteries, and at a much later date, an agricultural economy. The pottery and burial practices of the Dingsishan shell middens is identical to those found at the Da But sites of northern Vietnam, such as Da But, Con Co Ngua, Ban Ban Thuy, Lang Cong and Go Trung. These sites were occupied by complex hunter-gatherer populations during 5500–2000 BC (Viet 2007). Polished axes, pestles and mortars suggest cultivation, but clear evidence of food production only appears around 1500-1800 BC at sites such as Man Bac with domesticated pig remains (Sawada et al., 2011).
While this gives a picture of continuity between complex hunter-gatherers and agriculturalists in southern China and parts of northern Vietnam, elsewhere in mainland Southeast Asia continuity is harder to see. Hang Boi cave in inland northern Vietnam has a thick shell midden that spans just 10,600-12,300 BP (Rabett et al., 2011). At sites in Thailand there is a gap between cave occupation and open site occupation. At Lang Rongrien rockshelter on peninsular Thailand, the most recent dated occupation is about 8000 BP, followed by undated and highly disturbed deposits containing burials and pottery (Anderson, 1990:20). Similarly, in northern Thailand rockshelter occupation at Tham Lod and Ban Rai becomes discontinuous at around 8000 BP (Marwick and Gagan, 2011; Shoocongdej, 2006). At Laang Spean rockshelter in Cambodia, the most recent occupation in 5000 BP, followed by later disturbance of the stratigraphy (Sophady et al. 2015; Forestier 2015). The general pattern seems to be that cave and rockshelter sites switch from being occasional habitation sites to burial sites in the middle Holocene (Anderson 1997). A key challenge here is that the burials disturb the stratigraphy, making it difficult to assess continuity between forager occupation and later activity. At open air sites, the record starts at around 2000 BC (i.e. 4000 BP), for example at Khok Phanom Di, near the Bang Pakong River, southeast of Bangkok (Higham and Thosarat, 2004) and at Ban Non Wat (Higham and Kijngam, 2011). Occupation at these sites is characterized by burials, pottery, and in later phases, polished stone artefacts indicating crop cultivation.
To investigate the gap in the archaeological record between the shift from rockshelters to open sites during the middle Holocene, we chose to focus on coastal karstic valleys of Krabi province. This landscape has been exposed to major changes as sea levels rose and fell during the Late Pleistocene and Early Holocene (Sinsakul, 1992). This makes it well-suited to assessing the effects of local environmental change on forager groups during a time of major transitions in subsistence. Sinsakul (1992) has documented a Holocene sea level curve for Thailand that starts with a steady rise in sea level until about 6000 BP, reaching a height of 4 m amsl (above mean sea level) (Figure 1). Sea levels then regressed until 4700 BP, then rising again to 2.5 m amsl at about 4000 BP. From 3700 to 2700 BP there was a regressive phase, with transgression starting again at 2700 BP to a maximum of 2 m amsl at 2500 BP. Regression continued from that time until the present sea levels were reached at 1500 BP. The evidence for these sea level changes comes from direct dating of marine shells and peat deposits at geological sites in peninsular Thailand (Sinsakul, 1992).
Figure 1: Holocene sea level curve for Thailand, redrawn from Sinsakul (1992). Data points are radiocarbon ages of geological specimens from beach and tidal locations.
Previous research into archaeological correlates of these sea level changes in peninsula Thailand have been summarized by Anderson (2005). He describes faunal evidence from Lang Rongrien that has increases in marine shellfish around 7500 BP and between 4000 and 2500 BP. Anderson proposes that the increases in marine shellfish at the site are probably related to increases in sea levels. A small number of other sites have been previously investigated in several provinces of peninsula Thailand. For example, Moh Khiew in Krabi with its 25,000 year old human remains (Auetrakulvit et al., 2012, Chitkament, 2007; Matsumara and Pookajorn, 2005; Pookajorn, 1994), Tham Khao Khi Chan in Surat Thani Province has occupation layers dating from 6060 BP to 4250 BP (Srisuchat and Srisuchat, 1992). Buang Bap, also in Surat Thani, has faunal remains including marine shellfish dating between 6000 and 5000 BP (Srisuchat and Srisuchat, 1992). Pak Om has a dense and diverse archaeological deposit, but its two dates of 9350 and 3010 BP come from the same layer, so the chronology is uncertain (Srisuchat, 1997). Khao Tau in Pang Nga is a site complex with deep stratification and abundant cultural materials dating to 4750 and 5250 BP (Srisuchat and Srisuchat, 1992), Finally, there is the Tham Sua shell midden in Krabi that is a deposit of marine shell greater than one meter deep and with a basal date of 6440 BP (Anderson, 2005).
These previous excavations demonstrate human occupation at several sites in peninsular Thailand during the critical time of sea level changes in the Holocene. However, the level of available detail does not give a clear picture of stratigraphic integrity or the subsistence strategies represented at the sites. The goal of our work at Khao Toh Chong was to build on this previous work, not only by collecting an assemblage spanning the Holocene, but also by conducting geoarchaeological analyses at the site to assess stratigraphic integrity and provide local environmental context to the human occupation.
In June-July 2011 we excavated two areas of 2x2 m in 5 cm units to a depth of 1.6 m below the surface at Khao Toh Chong rockshelter. Excavated sediments were sieved using 5 mm and 10 mm screens. The site is a limestone overhang at the base of a 300 m high karst tower in Thap Prik Village. The rockshelter is about 30 m long with an average of about 10 m from the rear wall to the dripline. The dripline is about 40 m above the ground and a series of large boulders (3-4 m high) at the dripline give some protection from the wind and rain as well as trapping sediment in the shelter. The surface of the rockshelter is level fine sediment with no signs of disturbance and about 10 m above the surrounding ground, which is about 60 m above sea level. In Trench A, the southernmost trench, excavations reached a depth of 1.3 meters below the surface. In trench B, excavations were obstructed by bedrock in the northwest and southwest units. Subsequently, excavation depths in trench B extended to approximately 2.0 meters in the northeast and southeast units. Our archaeological and faunal analysis reported here is based on data from the southwest quadrant of trench A.
Bulk sediment samples collected from a column taken from the south wall of excavation trench A. A 1 g sub-sample was dried at 60°C for 24 hours for particle size analysis. The sub-sample was sieved to remove the >2 mm particles and carbonates were removed by washing the sample in 20 ml of 1 M HCl. Samples were then centrifuged and treated with 30 ml of 30% H2O2 for an hour to remove organics (Scott-Jackson and Walkington, 2005). Additional drying occurred for 30 hours in a 60°C oven. Each sample was added to a mixture of deionized water and surfactant Triton X 10 and agitated before being run in a Horiba LA-950 at the University of Washington Materials Science Department. A quartz refraction index of 1.458 was used during analysis.
We measured pH and electrical conductivity (EC) using a portable Oakton Waterproof Dual Parameter PCSTestr 35 on subsamples with a 1:1 ratio of sediment to deionized water. Soil organic material (SOM) and calcium carbonate content were measured by the Loss on Ignition method (after Gale and Hoare, 1991), as the percent of mass lost after heating samples to 600ºC for 4 hours and 1000ºC for 2 hours. Magnetic susceptibility was measured using a Bartington MS2 Magnetic Susceptibility Meter with 10 cm3 of sediment analyzed in sample pots at low and high frequency. Three replicates for each sample measurement of low and high frequency susceptibility were taken following Gale and Hoare (1991).
Samples for organic Carbon isotope analysis consisted of a 2 g sub-sample dried at 60°C for 24 hours, sieved to remove the >2mm particle size fraction (Hartman, 2011), with organics were picked out and discarded before samples were ground for 5 minutes using a mortar and pestle. To remove mineral carbonates in the sample, 60 mL of 1 M HCl was stirred into the samples and left to sit for 24 hours while stirring every 10 hours (Millwood and Boutton, 1998). To rinse the HCl off the samples, 60 mL of deionized water was stirred into the samples for 1 minute before setting to dry at 60°C for 48 h. Two more rinse cycles occurred where 60 mL DI water was stirred into the samples for one minute before setting to dry at 60°C for 24 hours each. Isotope measurement was conducted using a Costech Elemental Analyzer, Conflo III, MAT253 at the UW Earth and Space Sciences IsoLab.
For XRD analysis, samples were scanned on a Bruker D8 Focus X-Ray Diffractometer with a Cu radiation source. Following McGrath et al. (2008) we sub-sampled 2 g of >2 mm sediment and ground it to a fine powder. Next 20 ml of 30% H2O2 was used to remove organic matter. After effervescence it was removed and dried for another 60°C for 24 hours. After a final grinding, samples were loaded onto trays and scanned on a Bruker D8 Focus X-ray Diffractometer from 5° to 75° 2θ with a Cu radiation source at resolution 0.02° steps per second with 40 kV and 40 mA power output. MDI Jade 9 software was used to identify minerals.
Samples were prepared for compositional analysis by ICP-AES as follows. A 1 g sub-sample was added to 10ml of HNO3 and heated at 90°C for 15 minutes (Misarti et al., 2011). Another 5ml of HNO3 was next added and heated at 90°C for 60 minutes to drive the reaction. Next, deionized water and 30% H2O2 were added for increased oxidization and 10ml HCl was added and heated for chloride formation. The samples were diluted with deionized water and filtered before being measured and placed into Falcon tubes for ICP-AES analysis. This acid digest provides a broad spectrum of elements in a known volumetric concentration, as required for ICP-AES analysis (Balcerzak, 2002; Carter, 1993). The samples were analyzed in a Perkin Elmer Optima 8300DV in the University of Washington Chemistry Department.
We were unable to extract quantifiable amounts of pollen or phytoliths from the sediment samples (further details are reported in Van Vlack, 2014). This is likely due to the frequent wetting and drying of the rockshelter deposit which creates poor conditions for microflora preservation.
They key findings of our excavations were a faunal assemblage in a deposit with relatively few macroscopic traces of post-depositional disturbance (Figure 2). We did not encounter any burials or animal burrows and there is very limited termite activity in the deposit. We did not reach bedrock due to time constraints.
All excavated materials are currently stored at the Silpakorn University Faculty of Archaeology’s Phetchaburi campus. The raw data and code used to generate the results presented here have been organised into research compendium following the structure of an R package (Wickham, 2015) to enable reproducibility of the results (Marwick 2016). This compendium is archived online at DOI.
Figure 3: South section of Khao Toh Chong rockshelter trench A. The radiocarbon ages are the midpoints of the 95% calibrated age intervals. (c) indicates charcoal and (s) indicates shell as the material dated.